Logo of European Union
🧪 Open Loyalty AI Labs: From challenge to prototype in weeks
Learn more

Open Loyalty now speaks to AI agents

Open Loyalty MCP server is an open-source package that connects AI agents to your loyalty program
Table of contents
Recent update
June 26, 2026
Share this post

Every loyalty team we talk to is being asked the same question by their board this year. "Where is the AI?" Most of the honest answers sound the same too. The data sits in one system, the campaigns in another, and the loyalty engine speaks only to developers who know the API.

So we built the missing piece. Today we are releasing the Open Loyalty MCP server as an open-source package. It connects AI agents directly to your loyalty program, so the work that used to need a ticket and a developer can now happen in a conversation.

What MCP actually is

MCP, the Model Context Protocol, is a shared standard for letting AI assistants use real tools instead of just talking about them. Think of it as a universal adapter between an agent like Claude and a live system like Open Loyalty.

Because Open Loyalty has always been API-first, we were already most of the way there. The MCP server simply hands our API to the agent in a form it understands. We exposed 145 tools across 22 areas of the platform, from members and points to tiers, rewards, campaigns, segments, and analytics.

"We did not bolt AI onto loyalty. We made our loyalty engine something an agent can operate safely, end to end. That is a very different starting point, and it is only possible because the API was the product from day one."

Marcin Dyguda, Head of AI Innovation

What a loyalty team can do with it

The point is not the technology. It is the speed of the loop between an idea and a live change. A few examples from our own testing:

  • Ask which tier has the weakest retention this quarter, and get the answer pulled straight from the program.
  • Spin up a double-points campaign for members who have not purchased in 60 days, described in one sentence.
  • Build a segment, check a member balance, or issue a reward without opening the admin panel.
  • Draft and review a new challenge, then push it once a human approves.

A marketer gets a faster hand on the program. A developer gets hours back. The agent does the assembling, the person keeps the judgment.

Open by design

The server is published on npm as @open-loyalty/mcp-server under the MIT license. It runs locally with Claude Desktop, Claude Code, Cursor, and any MCP-compatible client. Connecting it takes one command and three settings.

This is an early exploration, and we are shipping it in the open on purpose. We would rather learn with the people building loyalty programs than guess in private. If your team is figuring out where AI belongs in your loyalty stack, this is a concrete place to start.

See the MCP server

Read the technical docs, or install it in your AI client in under a minute.

Read the docs

View on npm

API-first loyalty and gamification engine

Weekly tips to build & grow gamified loyalty programs
Join Loyalty Builders
About the authors
Marcin leads a team of loyalty experts who guide enterprises through the strategic design, implementation, and optimization of revenue-driving loyalty programs.
Join the community
of 4,000 Loyalty Builders!

Get a weekly dose of actionable tips on how to build and grow gamified successful loyalty programs!

Tell us about your challenges and we will together
Disney logo - blackMcDonald's logo - black

Customer loyalty know-how

Leverage resources from Open Loyalty’s gamification and loyalty experts to start smooth and move in the right direction